Research map / Probabilistic and scientific modeling / Variational and generative sampling
Efficient semidefinite-programming-based inference for binary and multi-class MRFs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use the linear-constraint SDP relaxation solved by the Mixing method | M^4 for multi-class MRF mode estimation · Synthetic multi-class MRFs, complete graphs, k=2,3,4,5 | relative error (mode): ~0.018 at worst | Helped |
| Use M^4 or M^4+ instead of AIS | Mode estimation · Synthetic complete graphs, k=2,3,4,5, CS(A)=2.5 | mode-estimation time | Helped |
| Estimate Z using M^4-guided rounding and importance sampling | Binary MRF partition-function estimation · Binary MRFs, complete and ER graphs | partition-function error (|log Z - log Z-hat|) | Helped |
| Use M^4-guided rounding and importance sampling instead of AIS | Multi-class MRF partition-function estimation · Multi-class MRFs, k>2 | partition-function accuracy and runtime | Helped |
| Use M^4+ rather than M^4 (ablation) | Partition-function estimation · Multi-class MRFs with larger k | partition-function accuracy | Hurt |
| Use M^4+ rather than M^4 (ablation) | Mode estimation · Synthetic multi-class MRFs, larger k | relative error (mode) | Hurt |
| Increase the number of AIS temperatures K (ablation) | AIS · Synthetic MRF partition-function estimation | partition-function accuracy | Helped |
| Increase the number of AIS temperatures K (ablation) | AIS · Synthetic MRF partition-function estimation | runtime | Hurt |
| Increase Gibbs-sampling cycles per AIS temperature (ablation) | AIS · Synthetic MRF partition-function estimation | partition-function accuracy | Helped |
| Increase Gibbs-sampling cycles per AIS temperature (ablation) | AIS · Synthetic MRF partition-function estimation | runtime | Hurt |
| Use randomized rounding samples to cover much of the probability mass | M^4-based partition-function estimator · 5-class MRF, CS(A)=2.5 | probability mass covered: 1000 iterations | Helped |
Closest papers by meaning
- Relaxation-Based Preprocessing Techniques for Markov Random Field Inference (CVPR 2016 · similarity 0.90)
- Inference in Graphical Models via Semidefinite Programming Hierarchies (NeurIPS 2017 · similarity 0.89)
- Continuous Relaxation of MAP Inference: A Nonconvex Perspective (CVPR 2018 · similarity 0.88)
- Lifted Hinge-Loss Markov Random Fields (AAAI 2019 · similarity 0.88)
- Inference in Higher Order MRF-MAP Problems With Small and Large Cliques (CVPR 2018 · similarity 0.88)
- The Linearization of Belief Propagation on Pairwise Markov Random Fields (AAAI 2017 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.